Analysing Qualitative Data: More Than ‘Identifying Themes’

Pat Bazeley

Research Support P/L and Australian Catholic University

pat@researchsupport.com.au

Final submitted copy: Malaysian Journal of Qualitative Research, 2009, 2, 6-22.

Abstract

Too often, qualitative researchers rely on the presentation of key themes supported by

quotes from participants’ text as the primary form of analysis and reporting of their data.

In this paper I argue that qualitative data require and support much deeper analysis.

Strategies that might assist researchers to enrich their analysis of qualitative data are

described. These strategies include improving interpretation and naming of categories;

using comparison and pattern analysis to refine and relate categories or themes; using

divergent views and negative cases to challenge generalisations; returning to substantive,

theoretical or methodological literature; creating displays using matrices, graphs, flow

charts and models; and using writing itself to prompt deeper thinking. Each strategy is

illustrated by examples.

Introduction: “Themes will be identified…”

Reliance on the identification of themes as the goal of analysis is endemic in qualitative

research. Often, in funding proposals, there will be a lengthy description of how

qualitative data are going to be gathered, but the only thing said about how these data are

going to be managed or analysed is that ‘themes will be identified in the data’. Similarly,

writers of journal articles often simply identify and discuss four or five ‘themes’ as their

analysis of the qualitative data in the study, with no attempt to link those themes into a

more comprehensive model of what they have found.

Researchers often use the terms concept, category and theme interchangeably in the

literature. I tend to use category for the descriptive level of coding and concept for a

more abstract level, and hence will often refer to categories and concepts when

discussing coding (Bazeley, 2007). Others (e.g., Strauss & Corbin, 1998) use concept for

the lower level, and category for a combination of several concepts. While theme is

sometimes used to describe an integrating, relational idea from the data (Richards, 2005),

more often it is used to describe elements identified from text and this is typically the

approach which is meant when people talk about identifying themes in the data as their

method of analysis.

Identifying themes has its place in qualitative research. They are a starting point in a

report of findings from a study. Effective reporting, however, requires your having used

data, and the ideas generated from the data, to build an argument that establishes the

point or points you wish to make. Strength of analysis will be recognised even by those

who may work differently, while descriptive reporting is likely to be unconvincing even

to those familiar with qualitative methods.

The Problem with ‘Themes’

Problems in ensuring quality analysis can begin from the point of research design and

data-making, through to the integration of data and drawing sound conclusions. For

example, if interview questions are asked in such a way as to prompt superficial answers,

there is not even a starting point for meaningful analysis. Further problems occur in

interpretation of data, in the ‘emergence’ and naming of themes, and in integrating

themes to provide a rich, deep understanding or a coordinated, explanatory model of what

has been found.

Interpreting Data

Once data are gathered, reading and interpretation are the starting points for meaningful

analysis. I have a problem, for example, with a researcher’s reading of this data segment:

He probably had been told that he had schizophrenia, but we didn’t

mention it, we didn’t say it, because it wasn’t our job to help him through

it. We felt it was the job of the psychiatric services that would have been

well trained in that, if he had been told by the doctor then he would know

what he faced.

This was interpreted as suggesting that satisfied participants had a clear idea of where

their caring responsibility ended and the responsibility of available professional mental

health services began. To me the far more significant issue raised was the silence around

mental illness, and the way in which such issues were avoided within the family. This

issue was never discussed by the researcher.

It can be very useful to share some small portions of your data with a colleague, or a

small group of colleagues (or fellow students) in a relaxed setting where individual and

collective attention is turned to making meaning from the passage being considered. This

has the benefit of serving as a reality check on your own interpretation, but more

particularly, discussion of that data segment typically creates added awareness of

dimensions in the data and prompts fresh ideas, with new questions to pursue.

Naming Themes and Connecting Data

Even where codes are appropriately developed, there is often a problem with the naming

of broader themes. Themes that are presented are often simply labels for metacategories

(a more inclusive category), or perhaps as a classification of codes into types of

categories. For example, Martin, Farrell, Lambrenos and Nayagam (2003) undertook

qualitative interviews with a sample of adolescents before and after their experience of

living with an Ilizarov frame—an external device fixed to a bone, designed to assist in the

lengthening of that bone over a period of up to nine months. They grouped codes into a

thematic structure which was thought to reflect common patterns emanating from the data

(Table 1). These ‘themes,’ however, amount to little more than a way of organising the

areas discussed by the adolescents. Because of the way they are named, they cannot

communicate to the reader without considerable explanation, and they cannot be used to

construct a meaningful summary or model of the experience. None of the themes was

examined comparatively in relation to age, gender, reason for needing the frame or

location of the frame. Nor were they related, on an individual basis, to the scaled

quantitative measures for coping or depression, despite this being presented as a mixed

methods study. No case histories were given to illustrate any connection of events (all

reports were group based).

Table 1: Themes emerging from qualitative analysis1

Before application of frame After application of frame

• An all-encompassing impact • Actuality of experience

• Coping resources • Coping and getting on

• Treatment expectations • Concerns, feelings and reactions

• Support for coping

• Advice and recommendations

• Treatment experiences 1

Originally presented as Table 4, in Martin et al. (2003, p. 483)

Additionally, in Martin et al.’s (2003) report, discrepancies between scaled measures and

interview data on the role of resignation and of social support as coping mechanisms

(both of which were more evident in the qualitative data) were not commented on until

discussion—and there the suggestion was simply that the children had come up with

aspects of these not thought of by the scale developers. Description is part of the analytic

journey, and ‘thick description’ is a valuable component of, for example,

phenomenological or ethnographic reporting, but description alone is not sufficient. The

data must be challenged, extended, supported, and linked in order to reveal their full

value.

Emergent Themes and Grounded Theory

In a draft report of a grounded theory study of what it means to be a good care worker,

the author wrote:

The interview data were initially coded according to a number of themes

that corresponded to the focus questions. Unanticipated issues and concerns

were raised and recurred in a number of interviews, for example, the care

workers repeatedly spoke of the importance of forming good relationships

to provide good care….

And,

A theme that recurred repeatedly was that central to being a good care

worker was having the ability to negotiate boundaries between personal and

professional relationships.

Were these themes of the importance of relationships and the need to maintain

boundaries really unanticipated? Are emergent themes actually emergent? The author had

just stated that the coding categories chosen corresponded to the focus questions, and

indeed, these questions (which were provided in an appendix) specifically asked about

relationships and boundaries.

Emergent themes are often remarkably similar to those in the literature (as indeed,

occurred also in this case). There is no problem with a priori categories or themes as long

as they are recognised and declared as such, and they are actually supported in the data;

the analyst can still retain flexibility and be open to the presence of finer nuances or

different emphases in the data. There is a problem, however, if something is written up as

unanticipated when it was clearly anticipated, and to have a point worth making in that

situation one does need to extend or relate the concept in fresh ways, to build new theory.

In this case, it could be helpful to ‘break open’ what relationship means when it is

between a care worker and an aged person, and so the work of analysis needs to shift in

that direction (cf. Figure 1, below).

Reporting Themes

These problems are further reflected in often shallow reporting in which themes are

typically presented using a brief summary and with a quote for each point as ‘evidence’

for the theme. There is a problem with relying for evidence on one or two quotes that

might have been drawn from hundreds of pages of text. While one or two quotes might

powerfully illustrate a theme, they do not convey how widely this theme might have

applied, or for whom, or how it links to other themes. Frequencies are sometimes

reported, but there is rarely any attempt to explain those who express this theme

differently, or who do not express the theme at all.

There is a problem also in being purely descriptive, presenting each theme in sequence,

just as there might be if the only report given from a survey was of simple frequencies or

means. Themes only attain full significance when they are linked to form a coordinated

picture or an explanatory model.

Moving from ‘Garden Path Analysis’ Toward a Coherent Model

My colleague, Lyn Richards, often talks about ‘garden path analysis’ when she is

teaching about qualitative analysis, as a way of showing how thematic ‘analysis’ can take

the reader along a pleasant pathway that leads nowhere: ‘Here are the roses, there are the

jonquils, and aren’t the daffodils lovely today!’ The suggestions which follow are

designed to help you move beyond the garden path toward a more meaningful and

coherent model or theory from your data.

Describe- Compare- Relate

‘Describe, compare, relate’ is a simple three-step formula I use when starting to work

through and record results of an analysis.

• Describing is an important starting point.

Outline the context for the study and provide details about sources of data, such as

the demographic features of the sample and the interrelationships between these

features. These give necessary background against which further analyses will be

read, as well as providing a basis for comparative analysis. Then move to the first

major category or ‘theme’. Describe (and record) its characteristics and

boundaries. How did people talk about this aspect, and how many talked about it?

What’s not included?

• Compare differences in the characteristics and boundaries for just that

category or theme across contrasting demographic groups or across

variations in context. Do themes occur more or less frequently for different groups? Are they expressed

differently by different groups? Ask questions of your data about this category or

theme—who, why, what, when? Record meaningful associations—doing so will

prompt further questions in your mind. Record, also, an absence of association—

not only is it important to know if there is no variation across groups or contexts,

recording these means you won’t need to waste time later re-checking.

• Relate this category or theme to others already written about. Ask more questions—does it make a difference if…? Use Strauss’ (1987) coding

paradigm to assist: Under what conditions does this category or theme arise?

What actions/interactions/strategies are involved? What are the consequences and

do these vary depending on the particular circumstances or the form in which it is

expressed? Record the questions you ask, and the results you find (or don’t find).

Repeat these three steps for each category or theme you want or need to write about. As

you relate categories you will be helped to structure your data because relating is best

done to categories already discussed. Thus, you will need to think about what the reader

already needs to know before they can understand what you are now writing about.

Parts of this initial approach to reporting the data will be preserved in the final report,

article or thesis, but further transformation will also occur. As you describe, compare and

relate for each element with an enquiring mind and an eye for evidence, your picture will

become increasingly complex and your theory or thesis will develop, building on the

foundation you have laid. Your analysis, then, will come together around an integrating

idea, with arguments to support it drawn from across your completed (interim) analyses.

Whether the integrating thesis/theory/model follows the description and analysis of key

themes as a conclusion drawn from them, or whether it is presented early and then

followed by further description will depend on the nature of the particular set of data and

the issues to be discussed.

In the draft of an article on what makes a good care worker, a doctoral student described

communication skills necessary for building and maintaining relationships between care

workers and clients in a way that made it very hard for the reader to take in (two quotes

followed, to support this extensive list):

Particular skills that were singled out were communication skills, skills in

negotiating and managing difficult situations, being able to take initiative

and be flexible, to exercise judgement, to be sensitive to and recognise

people’s needs and to deal with problems as they arose. In order to build a

working relationship with people, the participants stated they needed to be

able to listen carefully and respond appropriately to the client. They spoke

of needing highly developed ‘people skills’ that would enable care

workers to see things from the client’s point of view. They spoke of the

importance of being able to empathise and connect with people to be able

to talk and to listen when it was appropriate.

Exploring (from this passage only) how the various skills mentioned related to both

communication and to the goal of caring led to the following suggestion for an

alternative presentation:

Communication skills of care workers were demonstrated in two primary

ways (Figure 1):

• through their ability to negotiate and manage difficult situations, and

• through their capacity to listen and respond to the client.

Together, these skills reflect the practical and emotive dimensions of an

ability to take account of the client—a core category in what makes for

good care work.

The article might then continue with a description and analysis of each of these two

primary dimensions, but now in terms of how each involves communication skills and

contributes to the core process of taking account of the client.

Figure 1: The role of communication skills in ‘taking account of the client’

Using Divergent Views to Challenge Generalisations

Divergent views, negative cases or outliers—however you choose to label them—provide

a rich source for further analytic thinking, as you learn from them and grow your

understanding to incorporate them in your theorising (Miles & Huberman, 1994). In

qualitative work they cannot be ignored, but more than that, at times they provide the hint

that explains what is happening for the larger sample.

A doctoral student at a rural university in Australia is studying responses to

amalgamations of administrative regions within government health services in low-

density, rural areas. Communication issues were assuming a central focus in her analysis

(primarily, the lack of communication with employees throughout the process of

amalgamation). Two of her 20 interviewees stood out, however, in having positive views

about the amalgamations. So what was special about these two? One had taken a

redundancy and established a new career, another had been given a senior position in the

new system, with more power. Thus, those who were positive had personally benefited

from the amalgamations. This raised the question of whether those who were negative

were so more because they were now worse off in a personal sense than because the

service had deteriorated, and sent the student off on a new line of investigation to check

for evidence on selfish versus altruistic thinking in her interview and questionnaire data.

It challenged her generalisation about the centrality of the communication issue alone in

explaining interviewees’ responses, and eventually pointed to a much richer picture in

which communication issues, specifically lack of engagement and transparency by the

hierarchy, created a sense of uncertainty which fostered in employees a focus on their

own interests.

Work at proposing alternative explanations, of which negative cases are just one source,

then check the elements of these explanations or ideas generated from them against other

data (Yin, 2003). How widely are they supported? Can they be refined and developed?

Record these verifying strategies and their results, even if they prove to be false leads, as

this will help you build a case for your chosen explanation.

Returning to the Substantive, Theoretical and Methodological Literature

Your substantive and theoretical literature is another source of explanations to explore

and test. Read broadly. Your analysis may be stimulated by something from outside your

own area of work that nevertheless has application to it either in substance or in form. Or,

the way in which others have drawn from their data to reach an elegant and enlightening

conclusion may simply serve to inspire you to keep working at it!

Read the methodological literature for additional ideas on ways to refine or extend your

analysis strategies. For example, when I was writing about using NVivo for different

methodologies for the final chapter in Qualitative data analysis with NVivo (Bazeley,

2007), I read a number of chapters and articles on various forms of discourse analysis. I

make no pretence to having any expertise in discourse analysis, but the reading did at

least give me a new awareness of some features in my own data about academic

researchers, such that I could clearly identify at least three patterns of discourse in the

texts: discourses of performance, of romance, and of play. The danger is that, like

themes, these also can easily be presented as descriptive and disjointed observations and

so, like themes, they need to be integrated into a cohesive and purposeful analysis.

Creating and Using Displays from the Data

Miles and Huberman, in their classic 1994 text, argue cogently for the value of displaying

data to develop researcher understanding and for presentation of conclusions from the

data: “You know what you display” (p. 91). In displaying data, the researcher moves

from describing to explaining, through a “ladder of abstraction.” The form of a display

will vary depending on its purpose, the stage the analysis has reached, and on whether the

enquiry has a variable or process orientation.

Matrix displays: Whether they are drawn by hand or created through software, matrix

displays are an extremely useful way of detecting patterns in data.

The display in Figure 2 is part of a larger matrix that was used as an initial form of data

entry, designed to extract relevant information about research opportunities for new

academics from a series of interviews with 56 heads of departments across 6 disciplines

in 3 university types (Bazeley et al., 1996). From this display it was possible to compare

patterns for different departments and university types, facilitated by its having been

entered in Excel. For example, it became evident that while new staff in Physics and

Psychology were both supported in doing research (in contrast to some other disciplines),

they experienced considerable differences in teaching loads and thus in opportunities to

actually engage in research.

Figure 2: Matrix display of interview data, to examine patterns

Where text has been coded using software such as NVivo, those codes can be used to

construct matrix displays based on the co-occurrence of codes within the text, or of codes

and demographic attributes. The resulting matrix display (Figure 3) provides both the

frequency of responses and the detailed content of responses, allowing the researcher to

assess both patterns of association (how often things vary under different circumstances),

and the nature of the associations (in what ways something might vary under particular or

different circumstances).

Matrices are primarily useful for facilitating comparative analysis of data, and sometimes

for presenting conclusions.

Figure 3: Results from a matrix query in NVivo (Bazeley, 2007)

Flow charts and models are valuable early in a project to assist in initial conceptualisation

and planning, but their particular strength is as a means to present conclusions from an

analysis.

Figure 4 provides the overview for a more detailed series of flow charts through which I

presented my thesis that community development is an effective strategy for the

promotion of mental health in a disadvantaged population (Bazeley, 1977). What is

important here is that on the journey through my research I had already clarified what I

understood by key concepts for my research: a definition of mental health relevant to a

community context, the implications of disadvantage with respect to mental health, and

also my understanding of community development (which had been an unanticipated

component of my community mental health project). These clarifications greatly

facilitated my being able to visualise and theorise the links between community

development and mental health as I came to the conclusion of my project—indeed,

building the final model was possible only because I had undertaken these clarifications.

The way I often talk about this now with students is to ask them to write, at the end of

each segment of their work, what they are taking forward from it. Each component of

their writing then becomes focused, it provides a ‘road map’ for the reader, and each can

then contribute to building a conclusion.

Community development ���� Mental health

• Rebuilding community

Det

ails

of

spec

ific

lin

ks

foll

ow

ed i

n f

urt

her

ch

arts

• Avoid psychological

disorder

o shared identity � personal identity

� ‘whole person’ acceptance

� rootedness, commitment to

values

coping skills �

interpersonal support �

reduced stressors �

self esteem �

o absence of

impairment from

persistent

symptomatology

o mutual support � enhanced interpersonal

equity

� rootedness, security

� crisis management resources

• Personal/group

development through

decision making

• Self direction

o competence � personal coping skills

� environmental mastery

� individual instrumental skills

environmental mastery �

(competence)

o solve life problems

o power � personal

� societal

autonomy �

opportunity for varied �

roles and statuses

o make life choices

• The public context • Self actualisation

o social system linkages � participatory power

� increased knowledge

rootedness, commitment �

to values

‘whole person’ acceptance �

self esteem �

o personal integration

o instrumental changes in

the environment

� reduced environmental

stressors

� increased opportunities

development of skills �

opportunity for varied �

roles and statuses

o individuation

Figure 4: Community development for mental health in a disadvantaged community (Bazeley, 1977)

In her study of spinal injured people, Lynn Kemp (1999) developed a series of

increasingly complex theoretical models to illustrate various approaches to identifying

need for community services, including normative need, comparative need, perceived

need, felt need. Each was dismissed, in turn, as it was found to be not supported by her

data. At the end of the process, based on a combination of survey and interview data, she

reached an understanding of need as fulfilling a plan of life. For these spinal injured

people, the plan was to be ‘ordinary,’ but community services typically fostered

difference (Figure 5). Just occasionally, services supported the plan of the person with an

injury, but more often, in order to receive services, the recipient had to accept difference.

This model was built on careful consideration of the key themes in her participants’

narratives and, critically, the relationships between those themes, strongly supported by

the theoretical and research literature.

Figure 5: Plans of life and the role of services (Kemp, 1999)

Models for final conclusions and presentation purposes are often simplified from a more

complex version. I am reminded of the famous quote by Blaise Pascal: “The present letter

is a very long one, simply because I had no leisure to make it shorter.” Kemp’s model

was built on the basis of much more complex maps of interrelated categories but, in its

simple form, it holds within it a wealth of meaning that perfectly summed up both the

desires and the experiences of her spinal injured participants.

As well as providing a presentation tool, the process of creating a flow chart or model

will stimulate your thinking at any stage in the research process, as you determine how

the various elements (or themes) that you are investigating fit together. It was only as I

started to draw Figure 1, shown earlier, to puzzle about how the various listed categories

and themes might relate, that I realised that negotiation and listening/responding could be

seen as two kinds of communication and also two aspects of relationship, and that what

they had in common was the idea that each was a particular way of taking account of the

client. This model would inevitably change and develop as additional data and themes are

considered, but even as an interim step based on a small segment of summary text, it

Role of services

Spinal injured person

Ordinariness

Difference

proved to be enlightening and could begin to move the analysis beyond a simple

description of themes.

Typologies: A typology is a classification system built by viewing a concept along a

continuum, or perhaps by taking two (or more) dimensions to make an orthogonal display

(Patton, 2002). The logic of the pattern so made is then examined against the data to see

if this makes for meaningful subgroups, classes or ideal types, and for whether the pattern

holds in different settings. Labels for cells often draw on ‘indigenous’ terms, i.e. those

used by the participants in their everyday conversations. Patton describes a researcher’s

generation of fresh ideas through logically working between theory and data in this way

as ‘abductive thinking,’ while warning against manipulating the matrix in an attempt to

fill out all cells—all new ideas generated must be tested and confirmed by the data. In

that sense the typology created is a working tool, but possibly becomes also a final

presentation tool.

Table 2 is the initial result of my ‘playing around’ with cross-classifying some important

dimensions in researchers’ lives, derived from interview and focus group data. Once I

created it, I realised there must be at least one additional dimension needed (perhaps

opportunity?), as low support from the institution could equally well result in the

researcher’s experiencing moderate or high levels of frustration, leading to low research

activity—something that is not adequately captured in this table. I am also unsure about

some of the labels I’ve used. As a working model, nevertheless, it serves a purpose in

prompting further exploration of the relationships between personal and environmental

factors in research activity.

Table 2: A typology of academics’ responses to institutional research orientation

Research orientation of the institution

Low High

Level of

personal

commitment

to research

Low Non-researcher:

No need or opportunity

Reluctant researcher:

Only does what is required

Moderate Distracted researcher:

Always something more

important to do

Small ‘r’ researcher:

Engages in data gathering,

projects of local significance,

or with team

High Addicted researcher: Will do it anyway, but

may suffer in their

personal life as a

consequence

Big ‘R’ researcher: Harmony between personal

and institutional goals

creating possibility of major

interpretive breakthroughs

The creation of these various forms of display assumes that adequate concepts, categories

or themes are being (or have been) developed and substantiated in the available data.

Each has the potential to extend the analysis beyond those concepts, categories or themes,

to take the researcher into deeper understanding of experience or process, and hence to

theory building.

Writing Research as a Tool for Analysis

Like many others, I recommend that you start writing early, and that you keep writing.

Start Early

In qualitative research, the researcher’s reflective writing becomes a critical source of

interpretive understanding as concepts are dissected and ideas explored. Your reflective

writing, additionally, is invaluable in pointing to arguments to support your conclusions,

in that it provides an audit trail of how those conclusions were reached. Then, as you

move toward creating a report from your study, the exercise of writing in an ordered

presentation forces clarification of ideas.

As you start to write you will be prompted to go back to your data with further questions

to be resolved. There are practical benefits as well in beginning to write early: you don’t

become overwhelmed with data; you don’t become stuck with writer’s block—and you

don’t waste the world’s resources on printing out reams of unhelpful results for later

perusal.

Avoid Reliance on Quotes for Evidence

When the results of a qualitative project are presented as a series of themes, very often

each theme or sub-theme is presented as a statement followed by a quote designed to

provide evidence for the theme. Clearly, participants’ words must lie at the basis of the

conclusions you reach, but rarely will a participant make the argument for you in a few

words.

Reliance on presenting brief quoted segments of text as ‘evidence’ encourages superficial

reporting of themes, whereas building an argument requires that conclusions are drawn

from across the full range of available texts. One of the best strategies for ensuring you

write more than themes is to write the first draft of your results without any quotes. This

forces you to rely on wider evidence for what you are saying. Only when you have built

that evidence for your conclusions can you safely add some illustrative quotes to add

interest and clarity for the reader.

Before Lynn Kemp came up with her concluding model (Figure 5, above), she had made

two attempts to write results from the analysis of her interview data. In both she had

relied heavily on quotes; it was only when she was forced to write a version without

quotes that she saw and developed the central organising principle on which her final

thesis rested. She was then able to effectively illustrate that principle and demonstrate its

variations with quotes and selected case vignettes.

Keeping the Purpose in Focus

In recent work I have come across examples of researchers:

• writing to the sources – organising chapters by the kind of source (most

commonly, separating qualitative and quantitative sources);

• writing to the voices – organising chapters around the perspectives on the

topic held by each group of participants in a process;

• writing to the method – organising chapters around the approach taken to

analysis of the data where the particular method being used to guide the

analysis prescribed three different readings of the participants’ narratives.

This could also apply if different methodologies are employed in analysing

the same data (such as phenomenology and grounded theory).

In each of these cases the research purpose related to investigation of a substantive topic

and not to an analysis of the role of sources, voice or method. By organising their writing

around some aspect of the way in which they gained their data or approached their

analysis, rather than around what their data were saying with respect to the topic of the

research, the purpose was ‘hijacked’ and the text became repetitive. In each case that I

was involved with, shifting the organisation and focus of their writing to the topic of the

research brought fresh insights and a renewed sense of purpose to the writers as different

perspectives and data relating to each theme or issue could be brought together,

compared, contrasted and developed.

When Have You Arrived?

Lyn Richards (2005), in Handling qualitative data, suggests five signs of sufficiency for

an analysis:

• Simplicity – a ‘small polished gem of a theory’, rather than ‘a mere pebble of

truism;’

• Elegance and balance – it is coherent;

• Completeness – it explains all;

• Robustness – it doesn’t fall over with new data; and

• It makes sense to relevant audiences.

Conclusion

I have a Welsh Springer Spaniel dog. It is fascinating to watch him cross a field: he will

course back and forth in what appears to be a very indirect route until he picks up a

scent—his brain makes the connection with rabbit, and so then he will move rapidly and

directly in line with that scent. Arrival requires that you have moved along a tortuous and

possibly twisted path until you have found the scent, but having found it, you make the

connections and you are then able to lead the reader directly to the goal.

Identification of themes as a goal and as an end point of analysis fails Richards’ five

tests. Contextualising and making connections between those themes to build a coherent

argument supported by data is needed to satisfy immediate stakeholder audiences as well

as journal reviewers.

References

Bazeley, P. (1977). Community development for mental health. Unpublished doctoral

dissertation, Macquarie University.

Bazeley, P. (2007). Qualitative data analysis with NVivo. London: Sage.

Bazeley, P., Kemp, L., Stevens, K., Asmar, C., Grbich, C., Marsh, H., & Bhathal, R.

(1996). Waiting in the wings: a study of early career academic researchers in Australia.

National Board of Employment Education and Training, Commissioned Report No.50.

Canberra: Australian Government Publishing Service.

Kemp, L. A. (1999). Charting a parallel course: meeting the community service needs of

persons with spinal injuries. Unpublished doctoral dissertation, University of Western

Sydney.

Martin, L., Farrell, M., Lambrenos, K., & Nayagam, D. (2003). Living with the Ilizarov

frame: adolescent perceptions. Journal of Advanced Nursing, 43(5), 478-487.

Miles, M. B., & Huberman, A. M. (1994). Qualitative data analysis: An expanded

sourcebook. Thousand Oaks, CA: Sage.

Patton, M. Q. (2002). Qualitative evaluation and research methods (3rd ed.). Thousand

Oaks, CA: Sage.

Richards, L. (2005). Handling qualitative data. London: Sage.

Strauss, A., & Corbin, J. (1998). Basics of qualitative research (2nd ed.). Thousand Oaks,

CA: Sage.

Strauss, A. L. (1987). Qualitative analysis for social scientists. Cambridge, MA:

Cambridge University Press.

Yin, R. K. (2003). Case study research: design and methods (3rd ed.). Thousand Oaks,

CA: Sage.

About the Author:

Dr Pat Bazeley provides assistance and time out (and good food) to local and

international researchers from a wide range of disciplines at her research

retreat at Bowral, in the Southern Highlands of New South Wales. She also

holds senior, part-time appointments in Research Centres at the University of

New South Wales and at the Australian Catholic University, and has served as

an Associate Editor for the Journal of Mixed Methods Research. Her particular expertise

is in helping researchers to make sense of both quantitative and qualitative data and in

using computer software for management and analysis of data. Her publications focus on

qualitative and mixed methods data analysis, and on the development and performance of

researchers.